Peter Maydell [Mon, 30 Jan 2023 18:24:42 +0000 (18:24 +0000)]
target/arm: Make HSTR_EL2 traps take priority over UNDEF-at-EL1
The semantics of HSTR_EL2 require that it traps cpreg accesses
to EL2 for:
* EL1 accesses
* EL0 accesses, if the access is not UNDEFINED when the
trap bit is 0
(You can see this in the I_ZFGJP priority ordering, where HSTR_EL2
traps from EL1 to EL2 are priority 12, UNDEFs are priority 13, and
HSTR_EL2 traps from EL0 are priority 15.)
However, we don't get this right for EL1 accesses which UNDEF because
the register doesn't exist at all or because its ri->access bits
non-configurably forbid the access. At EL1, check for the HSTR_EL2
trap early, before either of these UNDEF reasons.
We have to retain the HSTR_EL2 check in access_check_cp_reg(),
because at EL0 any kind of UNDEF-to-EL1 (including "no such
register", "bad ri->access" and "ri->accessfn returns 'trap to EL1'")
takes precedence over the trap to EL2. But we only need to do that
check for EL0 now.
Signed-off-by: Peter Maydell <peter.maydell@linaro.org> Tested-by: Fuad Tabba <tabba@google.com> Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-id: 20230130182459.3309057-7-peter.maydell@linaro.org
Message-id: 20230127175507.2895013-7-peter.maydell@linaro.org
Peter Maydell [Mon, 30 Jan 2023 18:24:41 +0000 (18:24 +0000)]
target/arm: All UNDEF-at-EL0 traps take priority over HSTR_EL2 traps
The HSTR_EL2 register has a collection of trap bits which allow
trapping to EL2 for AArch32 EL0 or EL1 accesses to coprocessor
registers. The specification of these bits is that when the bit is
set we should trap
* EL1 accesses
* EL0 accesses, if the access is not UNDEFINED when the
trap bit is 0
In other words, all UNDEF traps from EL0 to EL1 take precedence over
the HSTR_EL2 trap to EL2. (Since this is all AArch32, the only kind
of trap-to-EL1 is the UNDEF.)
Our implementation doesn't quite get this right -- we check for traps
in the order:
* no such register
* ARMCPRegInfo::access bits
* HSTR_EL2 trap bits
* ARMCPRegInfo::accessfn
So UNDEFs that happen because of the access bits or because the
register doesn't exist at all correctly take priority over the
HSTR_EL2 trap, but where a register can UNDEF at EL0 because of the
accessfn we are incorrectly always taking the HSTR_EL2 trap. There
aren't many of these, but one example is the PMCR; if you look at the
access pseudocode for this register you can see that UNDEFs taken
because of the value of PMUSERENR.EN are checked before the HSTR_EL2
bit.
Rearrange helper_access_check_cp_reg() so that we always call the
accessfn, and use its return value if it indicates that the access
traps to EL0 rather than continuing to do the HSTR_EL2 check.
Signed-off-by: Peter Maydell <peter.maydell@linaro.org> Reviewed-by: Richard Henderson <richard.henderson@linaro.org> Tested-by: Fuad Tabba <tabba@google.com>
Message-id: 20230130182459.3309057-6-peter.maydell@linaro.org
Message-id: 20230127175507.2895013-6-peter.maydell@linaro.org
Rearrange the code in do_coproc_insn() so that we calculate the
syndrome value for a potential trap early; we're about to add a
second check that wants this value earlier than where it is currently
determined.
(Specifically, a trap to EL2 because of HSTR_EL2 should take
priority over an UNDEF to EL1, even when the UNDEF is because
the register does not exist at all or because its ri->access
bits non-configurably fail the access. So the check we put in
for HSTR_EL2 trapping at EL1 (which needs the syndrome) is
going to have to be done before the check "is the ARMCPRegInfo
pointer NULL".)
This commit is just code motion; the change to HSTR_EL2
handling that will use the 'syndrome' variable is in a
subsequent commit.
Signed-off-by: Peter Maydell <peter.maydell@linaro.org> Reviewed-by: Richard Henderson <richard.henderson@linaro.org> Tested-by: Fuad Tabba <tabba@google.com>
Message-id: 20230130182459.3309057-5-peter.maydell@linaro.org
Message-id: 20230127175507.2895013-5-peter.maydell@linaro.org
We added the CPAccessResult values CP_ACCESS_TRAP_UNCATEGORIZED_EL2
and CP_ACCESS_TRAP_UNCATEGORIZED_EL3 purely in order to use them in
the ats_access() function, but doing so was incorrect (a bug fixed in
a previous commit). There aren't any cases where we want an access
function to be able to request a trap to EL2 or EL3 with a zero
syndrome value, so remove these enum values.
As well as cleaning up dead code, the motivation here is that
we'd like to implement fine-grained-trap handling in
helper_access_check_cp_reg(). Although the fine-grained traps
to EL2 are always lower priority than trap-to-same-EL and
higher priority than trap-to-EL3, they are in the middle of
various other kinds of trap-to-EL2. Knowing that a trap-to-EL2
must always for us have the same syndrome (ie that an access
function will return CP_ACCESS_TRAP_EL2 and there is no other
kind of trap-to-EL2 enum value) means we don't have to try
to choose which of the two syndrome values to report if the
access would trap to EL2 both for the fine-grained-trap and
because the access function requires it.
Signed-off-by: Peter Maydell <peter.maydell@linaro.org> Reviewed-by: Richard Henderson <richard.henderson@linaro.org> Tested-by: Fuad Tabba <tabba@google.com>
Message-id: 20230130182459.3309057-4-peter.maydell@linaro.org
Message-id: 20230127175507.2895013-4-peter.maydell@linaro.org
Peter Maydell [Mon, 30 Jan 2023 18:24:38 +0000 (18:24 +0000)]
target/arm: Correct syndrome for ATS12NSO* at Secure EL1
The AArch32 ATS12NSO* address translation operations are supposed to
trap to either EL2 or EL3 if they're executed at Secure EL1 (which
can only happen if EL3 is AArch64). We implement this, but we got
the syndrome value wrong: like other traps to EL2 or EL3 on an
AArch32 cpreg access, they should report the 0x3 syndrome, not the
0x0 'uncategorized' syndrome. This is clear in the access pseudocode
for these instructions.
Fix the syndrome value for these operations by correcting the
returned value from the ats_access() function.
Signed-off-by: Peter Maydell <peter.maydell@linaro.org> Reviewed-by: Richard Henderson <richard.henderson@linaro.org> Tested-by: Fuad Tabba <tabba@google.com>
Message-id: 20230130182459.3309057-3-peter.maydell@linaro.org
Message-id: 20230127175507.2895013-3-peter.maydell@linaro.org
Peter Maydell [Mon, 30 Jan 2023 18:24:37 +0000 (18:24 +0000)]
target/arm: Name AT_S1E1RP and AT_S1E1WP cpregs correctly
The encodings 0,0,C7,C9,0 and 0,0,C7,C9,1 are AT SP1E1RP and AT
S1E1WP, but our ARMCPRegInfo definitions for them incorrectly name
them AT S1E1R and AT S1E1W (which are entirely different
instructions). Fix the names.
(This has no guest-visible effect as the names are for debug purposes
only.)
Signed-off-by: Peter Maydell <peter.maydell@linaro.org> Reviewed-by: Richard Henderson <richard.henderson@linaro.org> Tested-by: Fuad Tabba <tabba@google.com>
Message-id: 20230130182459.3309057-2-peter.maydell@linaro.org
Message-id: 20230127175507.2895013-2-peter.maydell@linaro.org
sbsa-ref: remove cortex-a76 from list of supported cpus
Cortex-A76 supports 40bits of address space. sbsa-ref's memory
starts above this limit.
Signed-off-by: Marcin Juszkiewicz <marcin.juszkiewicz@linaro.org> Reviewed-by: Alex Bennée <alex.bennee@linaro.org> Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-id: 20230126114416.2447685-1-marcin.juszkiewicz@linaro.org Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Alexander Graf [Fri, 23 Dec 2022 09:01:07 +0000 (10:01 +0100)]
hw/arm/virt: Make accels in GIC finalize logic explicit
Let's explicitly list out all accelerators that we support when trying to
determine the supported set of GIC versions. KVM was already separate, so
the only missing one is HVF which simply reuses all of TCG's emulation
code and thus has the same compatibility matrix.
Signed-off-by: Alexander Graf <agraf@csgraf.de> Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org> Reviewed-by: Cornelia Huck <cohuck@redhat.com> Reviewed-by: Zenghui Yu <yuzenghui@huawei.com> Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-id: 20221223090107.98888-3-agraf@csgraf.de
[PMM: Added qtest to the list of accelerators] Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Alexander Graf [Fri, 23 Dec 2022 09:01:06 +0000 (10:01 +0100)]
hw/arm/virt: Consolidate GIC finalize logic
Up to now, the finalize_gic_version() code open coded what is essentially
a support bitmap match between host/emulation environment and desired
target GIC type.
This open coding leads to undesirable side effects. For example, a VM with
KVM and -smp 10 will automatically choose GICv3 while the same command
line with TCG will stay on GICv2 and fail the launch.
This patch combines the TCG and KVM matching code paths by making
everything a 2 pass process. First, we determine which GIC versions the
current environment is able to support, then we go through a single
state machine to determine which target GIC mode that means for us.
After this patch, the only user noticable changes should be consolidated
error messages as well as TCG -M virt supporting -smp > 8 automatically.
Signed-off-by: Alexander Graf <agraf@csgraf.de> Reviewed-by: Richard Henderson <richard.henderson@linaro.org> Reviewed-by: Cornelia Huck <cohuck@redhat.com> Reviewed-by: Zenghui Yu <yuzenghui@huawei.com>
Message-id: 20221223090107.98888-2-agraf@csgraf.de Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Alexander Graf [Sat, 28 Jan 2023 22:44:59 +0000 (23:44 +0100)]
hvf: arm: Add support for GICv3
We currently only support GICv2 emulation. To also support GICv3, we will
need to pass a few system registers into their respective handler functions.
This patch adds support for HVF to call into the TCG callbacks for GICv3
system register handlers. This is safe because the GICv3 TCG code is generic
as long as we limit ourselves to EL0 and EL1 - which are the only modes
supported by HVF.
To make sure nobody trips over that, we also annotate callbacks that don't
work in HVF mode, such as EL state change hooks.
With GICv3 support in place, we can run with more than 8 vCPUs.
Signed-off-by: Alexander Graf <agraf@csgraf.de>
Message-id: 20230128224459.70676-1-agraf@csgraf.de Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Evgeny Iakovlev [Mon, 23 Jan 2023 16:23:03 +0000 (17:23 +0100)]
hw/char/pl011: better handling of FIFO flags on LCR reset
Current FIFO handling code does not reset RXFE/RXFF flags when guest
resets FIFO by writing to UARTLCR register, although internal FIFO state
is reset to 0 read count. Actual guest-visible flag update will happen
only on next data read or write attempt. As a result of that any guest
that expects RXFE flag to be set (and RXFF to be cleared) after resetting
FIFO will never see that happen.
Signed-off-by: Evgeny Iakovlev <eiakovlev@linux.microsoft.com> Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Message-id: 20230123162304.26254-5-eiakovlev@linux.microsoft.com Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Evgeny Iakovlev [Mon, 23 Jan 2023 16:23:01 +0000 (17:23 +0100)]
hw/char/pl011: add post_load hook for backwards-compatibility
Previous change slightly modified the way we handle data writes when
FIFO is disabled. Previously we kept incrementing read_pos and were
storing data at that position, although we only have a
single-register-deep FIFO now. Then we changed it to always store data
at pos 0.
If guest disables FIFO and the proceeds to read data, it will work out
fine, because we still read from current read_pos before setting it to
0.
However, to make code less fragile, introduce a post_load hook for
PL011State and move fixup read FIFO state when FIFO is disabled. Since
we are introducing a post_load hook, also do some sanity checking on
untrusted incoming input state.
Signed-off-by: Evgeny Iakovlev <eiakovlev@linux.microsoft.com>
Message-id: 20230123162304.26254-3-eiakovlev@linux.microsoft.com Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Evgeny Iakovlev [Mon, 23 Jan 2023 16:23:00 +0000 (17:23 +0100)]
hw/char/pl011: refactor FIFO depth handling code
PL011 can be in either of 2 modes depending guest config: FIFO and
single register. The last mode could be viewed as a 1-element-deep FIFO.
Current code open-codes a bunch of depth-dependent logic. Refactor FIFO
depth handling code to isolate calculating current FIFO depth.
One functional (albeit guest-invisible) side-effect of this change is
that previously we would always increment s->read_pos in UARTDR read
handler even if FIFO was disabled, now we are limiting read_pos to not
exceed FIFO depth (read_pos itself is reset to 0 if user disables FIFO).
Signed-off-by: Evgeny Iakovlev <eiakovlev@linux.microsoft.com> Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Message-id: 20230123162304.26254-2-eiakovlev@linux.microsoft.com Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Use the macro instead of two explicit string literals.
Signed-off-by: Richard Henderson <richard.henderson@linaro.org> Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org> Reviewed-by: Eric Auger <eric.auger@redhat.com>
Message-id: 20230124232059.4017615-1-richard.henderson@linaro.org Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Peter Maydell [Thu, 2 Feb 2023 18:00:40 +0000 (18:00 +0000)]
Merge tag 'pull-jan-omnibus-020223-1' of https://gitlab.com/stsquad/qemu into staging
Testing, docs, semihosting and plugin updates
- update playbooks for custom runners
- add section timing support to gitlab
- upgrade fedora images to 37
- purge perl from the build system and deps
- disable unstable tests in CI
- improve intro, emulation and semihosting docs
- semihosting bug fix and O_BINARY default
- add memory-sve test
- fix some races in qht
- improve plugin handling of memory helpers
- optimise plugin hooks
- fix some plugin deadlocks
- reduce win64-cross build time by dropping some targets
* tag 'pull-jan-omnibus-020223-1' of https://gitlab.com/stsquad/qemu: (36 commits)
gitlab: cut even more from cross-win64-system build
plugins: Iterate on cb_lists in qemu_plugin_user_exit
cpu-exec: assert that plugin_mem_cbs is NULL after execution
tcg: exclude non-memory effecting helpers from instrumentation
translator: always pair plugin_gen_insn_{start, end} calls
plugins: fix optimization in plugin_gen_disable_mem_helpers
plugins: make qemu_plugin_user_exit's locking order consistent with fork_start's
util/qht: use striped locks under TSAN
thread: de-const qemu_spin_destroy
util/qht: add missing atomic_set(hashes[i])
cpu: free cpu->tb_jmp_cache with RCU
tests/tcg: add memory-sve test for aarch64
semihosting: add O_BINARY flag in host_open for NT compatibility
semihosting: Write back semihosting data before completion callback
docs: add an introduction to the system docs
semihosting: add semihosting section to the docs
docs: add a new section to outline emulation support
docs: add hotlinks to about preface text
MAINTAINERS: Fix the entry for tests/tcg/nios2
gitlab: wrap up test results for custom runners
...
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Peter Maydell [Thu, 2 Feb 2023 13:42:56 +0000 (13:42 +0000)]
Merge tag 'pull-request-2023-01-31' of https://gitlab.com/thuth/qemu into staging
* qtest improvements
* Remove the deprecated OTP config of sifive_u
* Add libfdt to some of our CI jobs that were still missing it
* Use __builtin_bswap() everywhere (all compiler versions support it now)
* Deprecate the HAXM accelerator
* Document PCI devices handling on s390x
* Make Audiodev introspectable
* Improve the runtime of some CI jobs
* tag 'pull-request-2023-01-31' of https://gitlab.com/thuth/qemu: (27 commits)
gitlab-ci.d/buildtest: Merge the --without-default-* jobs
tests/qtest/display-vga-test: Add proper checks if a device is available
gitlab-ci.d/buildtest: Remove ppc-softmmu from the clang-system job
qapi, audio: Make introspection reflect build configuration more closely
qapi, audio: add query-audiodev command
docs/s390x/pcidevices: document pci devices on s390x
tests/qtest/boot-serial-test: Constify tests[] array
tests/qtest/vnc-display-test: Disable on Darwin
tests/qtest/vnc-display-test: Use the 'none' machine
tests/qtest/vnc-display-test: Suppress build warnings on Windows
tests/tcg: Do not build/run TCG tests if TCG is disabled
docs/about/deprecated: Mark HAXM in QEMU as deprecated
MAINTAINERS: Abort HAXM maintenance
qemu/bswap: Use compiler __builtin_bswap() on NetBSD
qemu/bswap: Use compiler __builtin_bswap() on FreeBSD
qemu/bswap: Use compiler __builtin_bswap() on Haiku
qemu/bswap: Remove <byteswap.h> dependency
qemu/bswap: Replace bswapXXs() by compiler __builtin_bswap()
qemu/bswap: Replace bswapXX() by compiler __builtin_bswap()
tests/docker/dockerfiles: Add libfdt to the i386 and to the riscv64 container
...
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Emilio Cota [Tue, 24 Jan 2023 18:01:25 +0000 (18:01 +0000)]
tcg: exclude non-memory effecting helpers from instrumentation
There are actually a whole bunch of helpers that don't affect memory
that we shouldn't instrument. They are helpfully identified by the
TCG_CALL_NO_SIDE_EFFECTS flag which marks out lookup_tb_ptr as well as
a lot of the maths helpers. To avoid the string compare we introduce a
new flag for plugin internals so we skip that too.
Related: #1381 Signed-off-by: Emilio Cota <cota@braap.org>
Message-Id: <20230108164731.61469-4-cota@braap.org>
[AJB: updated to skip all no SE plugins, add flag for plugin helper] Signed-off-by: Alex Bennée <alex.bennee@linaro.org> Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20230124180127.1881110-34-alex.bennee@linaro.org>
Emilio Cota [Tue, 24 Jan 2023 18:01:23 +0000 (18:01 +0000)]
plugins: fix optimization in plugin_gen_disable_mem_helpers
We were mistakenly checking tcg_ctx->plugin_insn as a canary to know
whether the TB had emitted helpers that might have accessed memory.
The problem is that tcg_ctx->plugin_insn gets updated on every
instruction in the TB, which results in us wrongly performing the
optimization (i.e. not clearing cpu->plugin_mem_cbs) way too often,
since it's not rare that the last instruction in the TB doesn't
use helpers.
Emilio Cota [Tue, 24 Jan 2023 18:01:18 +0000 (18:01 +0000)]
cpu: free cpu->tb_jmp_cache with RCU
Fixes the appended use-after-free. The root cause is that
during tb invalidation we use CPU_FOREACH, and therefore
to safely free a vCPU we must wait for an RCU grace period
to elapse.
$ x86_64-linux-user/qemu-x86_64 tests/tcg/x86_64-linux-user/munmap-pthread
=================================================================
==1800604==ERROR: AddressSanitizer: heap-use-after-free on address 0x62d0005f7418 at pc 0x5593da6704eb bp 0x7f4961a7ac70 sp 0x7f4961a7ac60
READ of size 8 at 0x62d0005f7418 thread T2
#0 0x5593da6704ea in tb_jmp_cache_inval_tb ../accel/tcg/tb-maint.c:244
#1 0x5593da6704ea in do_tb_phys_invalidate ../accel/tcg/tb-maint.c:290
#2 0x5593da670631 in tb_phys_invalidate__locked ../accel/tcg/tb-maint.c:306
#3 0x5593da670631 in tb_invalidate_phys_page_range__locked ../accel/tcg/tb-maint.c:542
#4 0x5593da67106d in tb_invalidate_phys_range ../accel/tcg/tb-maint.c:614
#5 0x5593da6a64d4 in target_munmap ../linux-user/mmap.c:766
#6 0x5593da6dba05 in do_syscall1 ../linux-user/syscall.c:10105
#7 0x5593da6f564c in do_syscall ../linux-user/syscall.c:13329
#8 0x5593da49e80c in cpu_loop ../linux-user/x86_64/../i386/cpu_loop.c:233
#9 0x5593da6be28c in clone_func ../linux-user/syscall.c:6633
#10 0x7f496231cb42 in start_thread nptl/pthread_create.c:442
#11 0x7f49623ae9ff (/lib/x86_64-linux-gnu/libc.so.6+0x1269ff)
0x62d0005f7418 is located 28696 bytes inside of 32768-byte region [0x62d0005f0400,0x62d0005f8400)
freed by thread T148 here:
#0 0x7f49627b6460 in __interceptor_free ../../../../src/libsanitizer/asan/asan_malloc_linux.cpp:52
#1 0x5593da5ac057 in cpu_exec_unrealizefn ../cpu.c:180
#2 0x5593da81f851 (/home/cota/src/qemu/build/qemu-x86_64+0x484851)
Signed-off-by: Emilio Cota <cota@braap.org> Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20230111151628.320011-2-cota@braap.org> Signed-off-by: Alex Bennée <alex.bennee@linaro.org>
Message-Id: <20230124180127.1881110-27-alex.bennee@linaro.org>
Alex Bennée [Tue, 24 Jan 2023 18:01:17 +0000 (18:01 +0000)]
tests/tcg: add memory-sve test for aarch64
This will be helpful in debugging problems with tracking SVE memory
accesses via the TCG plugins system.
Reviewed-by: Richard Henderson <richard.henderson@linaro.org> Signed-off-by: Alex Bennée <alex.bennee@linaro.org> Cc: Robert Henry <robhenry@microsoft.com> Cc: Aaron Lindsay <aaron@os.amperecomputing.com>
Message-Id: <20230124180127.1881110-26-alex.bennee@linaro.org>
Evgeny Iakovlev [Tue, 24 Jan 2023 18:01:16 +0000 (18:01 +0000)]
semihosting: add O_BINARY flag in host_open for NT compatibility
Windows open(2) implementation opens files in text mode by default and
needs a Windows-only O_BINARY flag to open files as binary. QEMU already
knows about that flag in osdep and it is defined to 0 on non-Windows,
so we can just add it to the host_flags for better compatibility.
Signed-off-by: Evgeny Iakovlev <eiakovlev@linux.microsoft.com> Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org> Reviewed-by: Bin Meng <bmeng.cn@gmail.com>
Message-Id: <20230106102018.20520-1-eiakovlev@linux.microsoft.com> Signed-off-by: Alex Bennée <alex.bennee@linaro.org>
Message-Id: <20230124180127.1881110-25-alex.bennee@linaro.org>
Keith Packard [Tue, 24 Jan 2023 18:01:15 +0000 (18:01 +0000)]
semihosting: Write back semihosting data before completion callback
'lock_user' allocates a host buffer to shadow a target buffer,
'unlock_user' copies that host buffer back to the target and frees the
host memory. If the completion function uses the target buffer, it
must be called after unlock_user to ensure the data are present.
This caused the arm-compatible TARGET_SYS_READC to fail as the
completion function, common_semi_readc_cb, pulled data from the target
buffer which would not have been gotten the console data.
I decided to fix all instances of this pattern instead of just the
console_read function to make things consistent and potentially fix
bugs in other cases.
Signed-off-by: Keith Packard <keithp@keithp.com> Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20221012014822.1242170-1-keithp@keithp.com> Signed-off-by: Alex Bennée <alex.bennee@linaro.org> Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Message-Id: <20230124180127.1881110-24-alex.bennee@linaro.org>
Alex Bennée [Tue, 24 Jan 2023 18:01:14 +0000 (18:01 +0000)]
docs: add an introduction to the system docs
Drop the frankly misleading quickstart section for a more rounded
introduction section. This new section gives an overview of the
accelerators as well as a high level introduction to some of the key
features of the emulator. We also expand on a general form for a QEMU
command line with a hopefully not too scary worked example of what
this looks like.
Acked-by: Richard Henderson <richard.henderson@linaro.org> Signed-off-by: Alex Bennée <alex.bennee@linaro.org> Reviewed-by: Kashyap Chamarthy <kchamart@redhat.com>
Message-Id: <20230124180127.1881110-23-alex.bennee@linaro.org>
Alex Bennée [Tue, 24 Jan 2023 18:01:13 +0000 (18:01 +0000)]
semihosting: add semihosting section to the docs
The main reason to do this is to document our O_BINARY implementation
decision somewhere. However I've also moved some of the implementation
details out of qemu-options and added links between the two. As a
bonus I've highlighted the scary warnings about host access with the
appropriate RST tags.
Acked-by: Richard Henderson <richard.henderson@linaro.org> Reviewed-by: Richard Henderson <richard.henderson@linaro.org> Signed-off-by: Alex Bennée <alex.bennee@linaro.org>
Message-Id: <20230124180127.1881110-22-alex.bennee@linaro.org>
Alex Bennée [Tue, 24 Jan 2023 18:01:12 +0000 (18:01 +0000)]
docs: add a new section to outline emulation support
This affects both system and user mode emulation so we should probably
list it up front.
Acked-by: Richard Henderson <richard.henderson@linaro.org> Signed-off-by: Alex Bennée <alex.bennee@linaro.org>
Message-Id: <20230124180127.1881110-21-alex.bennee@linaro.org>
Alex Bennée [Tue, 24 Jan 2023 18:01:11 +0000 (18:01 +0000)]
docs: add hotlinks to about preface text
Make it easier to navigate the documentation.
Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Acked-by: Richard Henderson <richard.henderson@linaro.org> Signed-off-by: Alex Bennée <alex.bennee@linaro.org> Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Message-Id: <20230124180127.1881110-20-alex.bennee@linaro.org>
Thomas Huth [Tue, 24 Jan 2023 18:01:10 +0000 (18:01 +0000)]
MAINTAINERS: Fix the entry for tests/tcg/nios2
tests/tcg/nios2/Makefile.target has accidentally been added
to the Microblaze section. Move it into the correct nios2
section instead - and while we're at it, it should also cover
the whole folder, and not only the Makefile.
Fixes: 67f80eb4d0 ("tests/tcg: enable debian-nios2-cross for test building") Signed-off-by: Thomas Huth <thuth@redhat.com> Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Message-Id: <20230119130326.2030297-1-thuth@redhat.com> Signed-off-by: Alex Bennée <alex.bennee@linaro.org>
Message-Id: <20230124180127.1881110-19-alex.bennee@linaro.org>
Alex Bennée [Tue, 24 Jan 2023 18:01:09 +0000 (18:01 +0000)]
gitlab: wrap up test results for custom runners
Instead of spewing the whole log to stdout lets just define them as
build artefacts so we can examine them later. Where we are running
check-tcg run it first as those tests are yet to be integrated into
meson. To avoid confusion we don't run multiple check-tcg tests at
once.
Reviewed-by: Thomas Huth <thuth@redhat.com> Signed-off-by: Alex Bennée <alex.bennee@linaro.org>
Message-Id: <20230124180127.1881110-18-alex.bennee@linaro.org>
linux-user blocks all signals while attempting to handle guest
signals (e.g. ABRT), which means that the default TERM sent by timeout
has no effect -- KILL instead.
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20230117035701.168514-2-richard.henderson@linaro.org>
[AJB: expanded commit message from cover letter] Signed-off-by: Alex Bennée <alex.bennee@linaro.org> Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Message-Id: <20230124180127.1881110-17-alex.bennee@linaro.org>
Alex Bennée [Tue, 24 Jan 2023 18:01:07 +0000 (18:01 +0000)]
tests/tcg: skip the vma-pthread test on CI
We are getting a lot of failures that are not related to changes so
this could be a flaky test.
Reviewed-by: Richard Henderson <richard.henderson@linaro.org> Signed-off-by: Alex Bennée <alex.bennee@linaro.org>
Message-Id: <20230124180127.1881110-16-alex.bennee@linaro.org>
Alex Bennée [Tue, 24 Jan 2023 18:01:06 +0000 (18:01 +0000)]
tests/docker: drop debian-tricore-cross's partial status
This image is perfectly capable of building QEMU, and indeed we do
that on gitlab. Drop the DOCKER_PARTIAL_IMAGES setting so we can also
test the gitlab build locally.
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org> Signed-off-by: Alex Bennée <alex.bennee@linaro.org>
Message-Id: <20230124180127.1881110-15-alex.bennee@linaro.org>
tests/docker: Install flex in debian-tricore-cross
When flex is not available, binutils sources default to the
'missing' script, but the current script available is not in
the format expected by the 'configure' script:
$ ./configure
...
/usr/src/binutils/missing: Unknown `--run' option
Try `/usr/src/binutils/missing --help' for more information
configure: WARNING: `missing' script is too old or missing
...
checking for bison... bison -y
checking for flex... no
checking for lex... no
checking for flex... /usr/src/binutils/missing flex
Alex Bennée [Tue, 24 Jan 2023 18:01:01 +0000 (18:01 +0000)]
gitlab: add lsan suppression file to workaround tcmalloc issues
The up-coming upgrade to Fedora 37 will bring in libtcmalloc as a
dependency of libglusterfs which confuses our fuzz run. Rather than
disable the build lets use LSAN's suppression mechanism to prevent the
job from failing.
Signed-off-by: Alex Bennée <alex.bennee@linaro.org> Cc: Daniel P. Berrangé <berrange@redhat.com> Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org> Reviewed-by: Thomas Huth <thuth@redhat.com>
Message-Id: <20230124180127.1881110-10-alex.bennee@linaro.org>
Alex Bennée [Tue, 24 Jan 2023 18:00:56 +0000 (18:00 +0000)]
tests/unit: drop hacky race avoidance in test-io-channel-command
We don't need to play timing games to ensure one socat wins over the
other, just create the fifo they both can use before spawning the
processes. However in the process we need to disable two tests for
Windows platforms as we don't have an abstraction for mkfifo().
Resolves: https://gitlab.com/qemu-project/qemu/-/issues/1403 Signed-off-by: Alex Bennée <alex.bennee@linaro.org> Reviewed-by: Thomas Huth <thuth@redhat.com>
Message-Id: <20230124180127.1881110-5-alex.bennee@linaro.org>
Alex Bennée [Tue, 24 Jan 2023 18:00:55 +0000 (18:00 +0000)]
gitlab: just use plain --cc=clang for custom runner build
I think this was because older Ubuntu's didn't alias clang to whatever
the latest version was. They do now so lets use that and not break.
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org> Reviewed-by: Richard Henderson <richard.henderson@linaro.org> Signed-off-by: Alex Bennée <alex.bennee@linaro.org>
Message-Id: <20230124180127.1881110-4-alex.bennee@linaro.org>
Mark Cave-Ayland [Tue, 24 Jan 2023 18:00:54 +0000 (18:00 +0000)]
gitlab: add FF_SCRIPT_SECTIONS for timings
Suggested-by: Mark Cave-Ayland <mark.cave-ayland@ilande.co.uk> Signed-off-by: Mark Cave-Ayland <mark.cave-ayland@ilande.co.uk> Signed-off-by: Alex Bennée <alex.bennee@linaro.org> Reviewed-by: Thomas Huth <thuth@redhat.com> Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org> Reviewed-by: Daniel P. Berrangé <berrange@redhat.com>
Message-Id: <20230124180127.1881110-3-alex.bennee@linaro.org>
Alex Bennée [Tue, 24 Jan 2023 18:00:53 +0000 (18:00 +0000)]
scripts/ci: update gitlab-runner playbook to use latest runner
We were using quite and old runner on our machines and running into
issues with stalling jobs. Gitlab in the meantime now reliably provide
the latest packaged versions of the runner under a stable URL. This
update:
- creates a per-arch subdir for builds
- switches from binary tarballs to deb packages
- re-uses the same binary for the secondary runner
- updates distro check for second to 22.04
Note this script isn't fully idempotent as we end up accumulating
runners especially during testing. However we also want to be able to
run twice with different GitLab keys (e.g. project and personal) so I
think we just have to be mindful of that during testing.
Signed-off-by: Alex Bennée <alex.bennee@linaro.org> Acked-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20230124180127.1881110-2-alex.bennee@linaro.org>
* tag 'python-pull-request' of https://gitlab.com/jsnow/qemu:
python/qemu/machine: use socketpair() for QMP by default
python/qmp/legacy: make QEMUMonitorProtocol accept a socket
python/qmp/protocol: add open_with_socket()
python/qmp: increase read buffer size
python/machine: Fix AF_UNIX path too long on macOS
python: QEMUMachine: enable qmp accept timeout by default
Fix some typos
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Thomas Huth [Mon, 30 Jan 2023 10:44:44 +0000 (11:44 +0100)]
gitlab-ci.d/buildtest: Merge the --without-default-* jobs
Let's safe some CI minutes by merging these two jobs. We can now
also drop "--disable-capstone" since the capstone submodule has
been removed a while ago. We should rather test --disable-fdt now
to check a compilation without the "dtc" submodule (for this we
have to drop i386-softmmu from the target list unfortunately).
Additionally, the qtests with s390x and sh4 are not read for
"--without-default-devices" yet, so we can only test mips64 and
avr here now.
Message-Id: <20230130104446.1286773-5-thuth@redhat.com> Reviewed-by: Alex Bennée <alex.bennee@linaro.org> Reviewed-by: Fabiano Rosas <farosas@suse.de> Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org> Signed-off-by: Thomas Huth <thuth@redhat.com>
Thomas Huth [Mon, 30 Jan 2023 10:44:43 +0000 (11:44 +0100)]
tests/qtest/display-vga-test: Add proper checks if a device is available
display-vga-test currently tries to guess the usable VGA devices
according to the target architecture that is used for the test.
This of course does not work if QEMU has been built with the
"--without-default-devices" configure switch. To fix this, use the
qtest_has_device() function for the decision instead. This way
we can also consolidate most of the test functions into one single
function (that takes a parameter with the device name now), except
for the multihead test that tries to instantiate two devices and
thus is a little bit different.
Message-Id: <20230130104446.1286773-4-thuth@redhat.com> Reviewed-by: Richard Henderson <richard.henderson@linaro.org> Signed-off-by: Thomas Huth <thuth@redhat.com>
Thomas Huth [Mon, 30 Jan 2023 10:44:41 +0000 (11:44 +0100)]
gitlab-ci.d/buildtest: Remove ppc-softmmu from the clang-system job
We are also compile-testing ppc64-softmmu with clang in the "tsan-build"
job, and ppc64-softmmu covers pretty much the same code as ppc-softmmu,
so we should not lose much test coverage here by removing ppc-softmmu
from the "clang-system" job.
Message-Id: <20230130104446.1286773-2-thuth@redhat.com> Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org> Signed-off-by: Thomas Huth <thuth@redhat.com>
qapi, audio: Make introspection reflect build configuration more closely
Currently the -audiodev accepts any audiodev type regardless of what is
built in to QEMU. An error only occurs later at runtime when a sound
device tries to use the audio backend.
With this change QEMU will immediately reject -audiodev args that are
not compiled into the binary. The QMP schema will also be introspectable
to identify what is compiled in.
This also helps to avoid compiling code that is not required in the
binary. Note: When building the audiodevs as modules, the patch only
compiles out code for modules that we don't build at all.
Signed-off-by: Daniel P. Berrangé <berrange@redhat.com>
[thuth: Rebase, take sndio and dbus devices into account]
Message-Id: <20230123083957.20349-3-thuth@redhat.com> Signed-off-by: Thomas Huth <thuth@redhat.com>
Way back in QEMU 4.0, the -audiodev command line option was introduced
for configuring audio backends. This CLI option does not use QemuOpts
so it is not visible for introspection in 'query-command-line-options',
instead using the QAPI Audiodev type. Unfortunately there is also no
QMP command that uses the Audiodev type, so it is not introspectable
with 'query-qmp-schema' either.
This introduces a 'query-audiodev' command that simply reflects back
the list of configured -audiodev command line options. This alone is
maybe not very useful by itself, but it makes Audiodev introspectable
via 'query-qmp-schema', so that libvirt (and other upper layer tools)
can discover the available audiodevs.
Signed-off-by: Daniel P. Berrangé <berrange@redhat.com>
[thuth: Update for upcoming QEMU v8.0, and use QAPI_LIST_PREPEND]
Message-Id: <20230123083957.20349-2-thuth@redhat.com> Signed-off-by: Thomas Huth <thuth@redhat.com>
Marcel Apfelbaum [Mon, 23 Jan 2023 17:42:05 +0000 (17:42 +0000)]
docs/pcie.txt: Replace ioh3420 with pcie-root-port
Do not mention ioh3420 in the "how to" doc.
The device still works and can be used by already
existing setups, but no need to be mentioned.
Suggested-by: Andrew Jones <drjones@redhat.com> Reviewed-by: Laszlo Ersek <lersek@redhat.com> Signed-off-by: Marcel Apfelbaum <marcel@redhat.com> Signed-off-by: Daniel P. Berrangé <berrange@redhat.com>
Message-Id: <20230123174205.683979-1-berrange@redhat.com> Reviewed-by: Michael S. Tsirkin <mst@redhat.com> Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
The nested event loop is broken by design. It's only user was removed.
Drop the code as well so that nobody ever tries to use it again.
I had to fix a couple of trivial conflicts around return values because
of 025faa872bcf ("vhost-user: stick to -errno error return convention").
Signed-off-by: Greg Kurz <groug@kaod.org>
Message-Id: <20230119172424.478268-3-groug@kaod.org> Reviewed-by: Michael S. Tsirkin <mst@redhat.com> Signed-off-by: Michael S. Tsirkin <mst@redhat.com> Acked-by: Maxime Coquelin <maxime.coquelin@redhat.com>
Motivation : this is breaking vhost-user with DPDK as reported in [0].
Received unexpected msg type. Expected 22 received 40
Fail to update device iotlb
Received unexpected msg type. Expected 40 received 22
Received unexpected msg type. Expected 22 received 11
Fail to update device iotlb
Received unexpected msg type. Expected 11 received 22
vhost VQ 1 ring restore failed: -71: Protocol error (71)
Received unexpected msg type. Expected 22 received 11
Fail to update device iotlb
Received unexpected msg type. Expected 11 received 22
vhost VQ 0 ring restore failed: -71: Protocol error (71)
unable to start vhost net: 71: falling back on userspace virtio
The failing sequence that leads to the first error is :
- QEMU sends a VHOST_USER_GET_STATUS (40) request to DPDK on the master
socket
- QEMU starts a nested event loop in order to wait for the
VHOST_USER_GET_STATUS response and to be able to process messages from
the slave channel
- DPDK sends a couple of legitimate IOTLB miss messages on the slave
channel
- QEMU processes each IOTLB request and sends VHOST_USER_IOTLB_MSG (22)
updates on the master socket
- QEMU assumes to receive a response for the latest VHOST_USER_IOTLB_MSG
but it gets the response for the VHOST_USER_GET_STATUS instead
The subsequent errors have the same root cause : the nested event loop
breaks the order by design. It lures QEMU to expect responses to the
latest message sent on the master socket to arrive first.
Since this was only needed for DAX enablement which is still not merged
upstream, just drop the code for now. A working solution will have to
be merged later on. Likely protect the master socket with a mutex
and service the slave channel with a separate thread, as discussed with
Maxime in the mail thread below.
Reported-by: Yanghang Liu <yanghliu@redhat.com> Buglink: https://bugzilla.redhat.com/show_bug.cgi?id=2155173 Signed-off-by: Greg Kurz <groug@kaod.org>
Message-Id: <20230119172424.478268-2-groug@kaod.org> Reviewed-by: Michael S. Tsirkin <mst@redhat.com> Signed-off-by: Michael S. Tsirkin <mst@redhat.com> Acked-by: Stefan Hajnoczi <stefanha@redhat.com> Acked-by: Maxime Coquelin <maxime.coquelin@redhat.com>
Thomas Huth [Wed, 18 Jan 2023 12:51:32 +0000 (13:51 +0100)]
tests/qtest/bios-tables-test: Make the test less verbose by default
We are facing the issues that our test logs in the gitlab CI are
too big (and thus cut off). The bios-tables-test is one of the few
qtests that prints many lines of output by default when running with
V=1, so it contributes to this problem. Almost all other qtests are
silent with V=1 and only print debug messages with V=2 and higher.
Thus let's change the bios-tables-test to behave more like the
other tests and only print the debug messages with V=2 (or higher).
Signed-off-by: Thomas Huth <thuth@redhat.com>
Message-Id: <20230118125132.1694469-1-thuth@redhat.com> Reviewed-by: Michael S. Tsirkin <mst@redhat.com> Signed-off-by: Michael S. Tsirkin <mst@redhat.com> Reviewed-by: Daniel P. Berrangé <berrange@redhat.com> Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org> Reviewed-by: Igor Mammedov <imammedo@redhat.com>
Use the proper QOM type definition instead of magic string.
This also helps during eventual refactor while using git-grep.
Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Message-Id: <20230117193014.83502-1-philmd@linaro.org> Reviewed-by: Michael S. Tsirkin <mst@redhat.com> Signed-off-by: Michael S. Tsirkin <mst@redhat.com> Reviewed-by: Eric Auger <eric.auger@redhat.com>
The VHOST_USER_ADD/REM_MEM_REG requests should be categorized into
non-vring specific messages, and should be sent only once.
Signed-off-by: Minghao Yuan <yuanmh12@chinatelecom.cn>
Message-Id: <20230123122119.194347-1-yuanmh12@chinatelecom.cn> Reviewed-by: Michael S. Tsirkin <mst@redhat.com> Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Igor Mammedov [Thu, 12 Jan 2023 14:03:11 +0000 (15:03 +0100)]
pcihp: generate populated non-hotpluggble slot descriptions on non-hotplug path
Generating slots descriptions populated by non-hotpluggable devices
is akward at best and complicates hotplug path (build_append_pcihp_slots)
needlessly, and builds only dynamic _DSM for such slots which is overlkill.
Clean it up and let non-hotplug path (build_append_pci_bus_devices)
to handle that task.
Such clean up effectively drops dynamic _DSM methods on non-hotpluggable
slots (even though bus itself is hotpluggable), but in practice it
affects only built-in devices (ide controllers/various bridges) that don't
use acpi-index anyways so effectively it doesn't matter (NICs are hotpluggble).
Follow up series will add static _DSM for non-hotpluggble devices/buses
that will not depend on ACPI PCI hotplug at all, and potentially would
allows us to reuse non-hotplug path elsewhere (PBX/microvm/arm-virt),
including new support for acpi-index for non-hotpluggable devices.
Signed-off-by: Igor Mammedov <imammedo@redhat.com>
Message-Id: <20230112140312.3096331-40-imammedo@redhat.com> Reviewed-by: Michael S. Tsirkin <mst@redhat.com> Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Igor Mammedov [Thu, 12 Jan 2023 14:03:10 +0000 (15:03 +0100)]
tests: acpi: whitelist DSDT before moving non-hotpluggble slots description from hotplug path
Signed-off-by: Igor Mammedov <imammedo@redhat.com>
Message-Id: <20230112140312.3096331-39-imammedo@redhat.com> Reviewed-by: Michael S. Tsirkin <mst@redhat.com> Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Igor Mammedov [Thu, 12 Jan 2023 14:03:08 +0000 (15:03 +0100)]
pcihp: acpi: ignore coldplugged bridges when composing hotpluggable slots
coldplugged bridges are not unpluggable, so there is no need
to describe slots where they are plugged as hotpluggable. To
that effect we have a condition that marks slot as non-hotpluggable
if it's populated by coldplugged bridge and prevents generation
_SUN/_EJ0 objects for it. That leaves dynamic _DSM method on
such slot (which also depends on BSEL and pcihp hardware).
This _DSM method provides only dynamic acpi-index support so far,
which is not actually used/supported by linux kernel for bridges
and it's doubtful there will be need for it at all.
So it's rather pointless to generate acpi-index related AML
for bridges and we can simplify hotplug slots generator a bit
more by completely ignoring coldplugged bridges on hotplug path.
Another point in favor of dropping dynamic _DSM support, is
that we can replace it with static _DSM if necessary since
a slot with bridge can't change during VM runtime and without
any dependency on ACPI PCI hotplug at that.
Later I plan to implement bridge specific static _DSM
PCI Firmware Specification 3.2
4.6.5. _DSM for Ignoring PCI Boot Configurations
part of spec, to fix longstanding issue with fixed IO/MEM
resource assignment that often leads to hotplugged device
being in-operational within the guest due limited IO/MEM
windows programmed on bridge at boot time.
Signed-off-by: Igor Mammedov <imammedo@redhat.com>
Message-Id: <20230112140312.3096331-37-imammedo@redhat.com> Reviewed-by: Michael S. Tsirkin <mst@redhat.com> Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Igor Mammedov [Thu, 12 Jan 2023 14:03:07 +0000 (15:03 +0100)]
tests: acpi: whitelist DSDT blobs before removing dynamic _DSM on coldplugged bridges
Signed-off-by: Igor Mammedov <imammedo@redhat.com>
Message-Id: <20230112140312.3096331-36-imammedo@redhat.com> Reviewed-by: Michael S. Tsirkin <mst@redhat.com> Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Igor Mammedov [Thu, 12 Jan 2023 14:03:05 +0000 (15:03 +0100)]
pcihp: acpi: decouple hotplug and generic slots description
Split build_append_pci_bus_devices() onto generic part that builds
AML descriptions only for populated slots which is applicable to
both hotplug disabled and enabled bridges. And a hotplug only
part that complements generic AML with hotplug depended bits
(that depend on BSEL), like _SUN/_EJ0 entries, dynamic _DSM.
Hotplug part, will generate full 'Device' descriptors for
non-populated slots (like it used to be) and complementary
'Scope' descriptors for populated slots that are hotplug capable.
i.e. something like this:
- ...
+ Name (BSEL, 0x03)
+ Scope (S00)
+ {
+ Name (ASUN, Zero)
+ Method (_DSM, 4, Serialized) // _DSM: Device-Specific Method
+ {
+ Local0 = Package (0x02)
+ {
+ BSEL,
+ ASUN
+ }
+ Return (PDSM (Arg0, Arg1, Arg2, Arg3, Local0))
+ }
+ [ ... other hotplug depended bits ]
+ }
While generic build_append_pci_bus_devices() still calls hotplug part at
its end it doesn't really depend on any hotplug bits anymore and later
both could be completely separated when it's necessary.
Main benefit though is that both build_append_pci_bus_devices() and
build_append_pcihp_slots() become more readable and it makes easier
to modify them with less risk of affecting another part. Also it opens
possibility to re-use generic part elsewhere (microvm, arm/virt).
Signed-off-by: Igor Mammedov <imammedo@redhat.com>
Message-Id: <20230112140312.3096331-34-imammedo@redhat.com> Reviewed-by: Michael S. Tsirkin <mst@redhat.com> Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Igor Mammedov [Thu, 12 Jan 2023 14:03:04 +0000 (15:03 +0100)]
tests: acpi: whitelist DSDT before decoupling PCI hotplug code from basic slots description
Signed-off-by: Igor Mammedov <imammedo@redhat.com>
Message-Id: <20230112140312.3096331-33-imammedo@redhat.com> Reviewed-by: Michael S. Tsirkin <mst@redhat.com> Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Igor Mammedov [Thu, 12 Jan 2023 14:03:03 +0000 (15:03 +0100)]
pcihp: isolate rule whether slot should be described in DSDT
Signed-off-by: Igor Mammedov <imammedo@redhat.com>
Message-Id: <20230112140312.3096331-32-imammedo@redhat.com> Reviewed-by: Michael S. Tsirkin <mst@redhat.com> Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Igor Mammedov [Thu, 12 Jan 2023 14:03:02 +0000 (15:03 +0100)]
pci: make sure pci_bus_is_express() won't error out with "discards ‘const’ qualifier"
function doesn't need RW aceess to passed in bus pointer,
make it const.
Signed-off-by: Igor Mammedov <imammedo@redhat.com>
Message-Id: <20230112140312.3096331-31-imammedo@redhat.com> Reviewed-by: Michael S. Tsirkin <mst@redhat.com> Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Igor Mammedov [Thu, 12 Jan 2023 14:03:01 +0000 (15:03 +0100)]
pcihp: make bridge describe itself using AcpiDevAmlIfClass:build_dev_aml
simplify build_append_pci_bus_devices() a bit by handling bridge
specific logic in bridge dedicated AcpiDevAmlIfClass::build_dev_aml
callback.
Signed-off-by: Igor Mammedov <imammedo@redhat.com>
Message-Id: <20230112140312.3096331-30-imammedo@redhat.com> Reviewed-by: Michael S. Tsirkin <mst@redhat.com> Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Igor Mammedov [Thu, 12 Jan 2023 14:03:00 +0000 (15:03 +0100)]
pci: acpi: wire up AcpiDevAmlIf interface to generic bridge
... so that the concrete impl. won't has to duplicate it
every time. By default it doesn't do anything unless leaf class
defines and sets AcpiDevAmlIfClass::build_dev_aml handler.
Signed-off-by: Igor Mammedov <imammedo@redhat.com>
Message-Id: <20230112140312.3096331-29-imammedo@redhat.com> Reviewed-by: Michael S. Tsirkin <mst@redhat.com> Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Igor Mammedov [Thu, 12 Jan 2023 14:02:59 +0000 (15:02 +0100)]
x86: pcihp: acpi: prepare slot ignore rule to work with self describing bridges
Before switching pci bridges to AcpiDevAmlIf interface, ensure that
ignored slots are handled correctly.
(existing rule works but only if bridge doesn't have AcpiDevAmlIf interface).
While at it rewrite related comments to be less confusing (hopefully).
Signed-off-by: Igor Mammedov <imammedo@redhat.com>
Message-Id: <20230112140312.3096331-28-imammedo@redhat.com> Reviewed-by: Michael S. Tsirkin <mst@redhat.com> Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Igor Mammedov [Thu, 12 Jan 2023 14:02:58 +0000 (15:02 +0100)]
tests: acpi: update expected blobs
previous commit added endpoint devices to bridge testcases,
which exposes extra non-hotpluggable slot in DSDT on bus where
hotplug is not available.
It should look like this (numbers may vary):
Igor Mammedov [Thu, 12 Jan 2023 14:02:57 +0000 (15:02 +0100)]
tests: acpi: add endpoint devices to bridges
to make sure that they are enumerated or ignored as expected
Signed-off-by: Igor Mammedov <imammedo@redhat.com>
Message-Id: <20230112140312.3096331-26-imammedo@redhat.com> Reviewed-by: Michael S. Tsirkin <mst@redhat.com> Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Igor Mammedov [Thu, 12 Jan 2023 14:02:56 +0000 (15:02 +0100)]
whitelist DSDT before adding endpoint devices to bridge testcases
Signed-off-by: Igor Mammedov <imammedo@redhat.com>
Message-Id: <20230112140312.3096331-25-imammedo@redhat.com> Reviewed-by: Michael S. Tsirkin <mst@redhat.com> Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
* due to ("pcihp: do not put empty PCNT in DSDT") in the most Q35 tests
...
{
Name (_ADR, 0x001F0003) // _ADR: Address
}
-
- Method (PCNT, 0, NotSerialized)
- {
- }
}
}
Igor Mammedov [Thu, 12 Jan 2023 14:02:54 +0000 (15:02 +0100)]
pcihp: do not put empty PCNT in DSDT
count number of PCNT methods that actually call Notify
and if there aren't any, drop PCNT altogether.
It mostly affects 'Q35' tests where there is no root-ports
/bridges attached and 'PC' machine when ACPI PCI hotplug is
completely disabled.
Signed-off-by: Igor Mammedov <imammedo@redhat.com>
Message-Id: <20230112140312.3096331-23-imammedo@redhat.com> Reviewed-by: Michael S. Tsirkin <mst@redhat.com> Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Igor Mammedov [Thu, 12 Jan 2023 14:02:53 +0000 (15:02 +0100)]
pcihp: compose PCNT callchain right before its user _GPE._E01
it's a stepping stone to making build_append_pci_bus_devices() suitable
for AcpiDevAmlIfClass:build_dev_aml callback and lets further simplify
it by separating PCNT generation from slots descriptions.
It also makes PCNT callchain ASL much more readable since callchain
not longer cluttered by slots descriptors.
Plus, move will let next patch easily drop empty PCNT (pc/q35)
when there is nothing hotpluggable.
Signed-off-by: Igor Mammedov <imammedo@redhat.com>
Message-Id: <20230112140312.3096331-22-imammedo@redhat.com> Reviewed-by: Michael S. Tsirkin <mst@redhat.com> Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Igor Mammedov [Thu, 12 Jan 2023 14:02:50 +0000 (15:02 +0100)]
pcihp: drop pcihp_bridge_en dependency when composing PCNT method
.. and use only BSEL presence to decide on how PCNT should be composed.
That simplifies possible combinations to consider, but mainly it makes
PCIHP AML be governed only by BSEL, which is property of PCIBus
(aka part of bridge) and as result it opens possibility to convert
build_append_pci_bus_devices() into AcpiDevAmlIf::build_dev_aml
callback to make bridges self describing.
PS:
used approach leaves unused PCNT, when ACPI hotplug is completely
disabled but that's harmless and followup commits will get rid of
it later.
Signed-off-by: Igor Mammedov <imammedo@redhat.com>
Message-Id: <20230112140312.3096331-19-imammedo@redhat.com> Reviewed-by: Michael S. Tsirkin <mst@redhat.com> Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Igor Mammedov [Thu, 12 Jan 2023 14:02:49 +0000 (15:02 +0100)]
tests: acpi: whitelist DSDT before refactoring acpi based PCI hotplug machinery
Signed-off-by: Igor Mammedov <imammedo@redhat.com>
Message-Id: <20230112140312.3096331-18-imammedo@redhat.com> Reviewed-by: Michael S. Tsirkin <mst@redhat.com> Signed-off-by: Michael S. Tsirkin <mst@redhat.com> Reviewed-by: Michael S. Tsirkin <mst@redhat.com> Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Igor Mammedov [Thu, 12 Jan 2023 14:02:48 +0000 (15:02 +0100)]
tests: acpi: add reboot cycle to bridge test
hotplugged bridges should not be described in DSDT,
while it works on cold boot, some ACPPI PCI code
are invoked during reboot.
This patch will let us catch unexpected AML if hotplug
checks are broken.
Signed-off-by: Igor Mammedov <imammedo@redhat.com>
Message-Id: <20230112140312.3096331-17-imammedo@redhat.com> Reviewed-by: Michael S. Tsirkin <mst@redhat.com> Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Igor Mammedov [Thu, 12 Jan 2023 14:02:47 +0000 (15:02 +0100)]
tests: boot_sector_test(): make it multi-shot
if the function is called the 2nd time within the same qtest session,
it will prematurely return before boot sector is executed due to
remaining signature.
Follow up patch will add VM reboot to a test case and will
call boot_sector_test() again within the same qtest env,
which may lead to above issue.
To fix it make sure signature in VM RAM is no more before
exiting boot_sector_test(), so next time it's called it
will wait boot sector is completed again.
Signed-off-by: Igor Mammedov <imammedo@redhat.com>
Message-Id: <20230112140312.3096331-16-imammedo@redhat.com> Reviewed-by: Michael S. Tsirkin <mst@redhat.com> Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Igor Mammedov [Thu, 12 Jan 2023 14:02:46 +0000 (15:02 +0100)]
tests: acpi: extend bridge tests with hotplugged bridges
with previous commit fixing malformed PCNT calls to hotplugged
bridges, it should be possible add coldplug/hotplug test when
describing PCI topology in DSDT without breeaking CI.
Signed-off-by: Igor Mammedov <imammedo@redhat.com>
Message-Id: <20230112140312.3096331-15-imammedo@redhat.com> Reviewed-by: Michael S. Tsirkin <mst@redhat.com> Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Igor Mammedov [Thu, 12 Jan 2023 14:02:45 +0000 (15:02 +0100)]
tests: boot_sector_test: avoid crashing if status is not available yet
If test case was started in paused mode (-S CLI option) and then
allowed to continue via QMP, boot_sector_test could assert on
transient state with following error:
Instead of crashing test if 'status' is not available yet, skip check
and repeat iteration again after TEST_DELAY has elapsed.
Signed-off-by: Igor Mammedov <imammedo@redhat.com>
Message-Id: <20230112140312.3096331-14-imammedo@redhat.com> Reviewed-by: Michael S. Tsirkin <mst@redhat.com> Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
it will generate AML calls to hpbr's PCNT, which doesn't exists
since it's hotplugged bridge. As result DSDT becomes malformed,
with consequences that hotplug might stop working at best or
crash guest OS at worst, when it attempts to call non existing
PCNT method or during OS guest reboot when parsing DSDT again.
IASL de-compiles malformed AML of above config DSDT as:
With BSEL assignment limited only to coldplugged bridges [1],
it should be possible to add PCNT call to a child bridge only
if the child has BSEL property, otherwise ignore it since it's
hotplugged. Which should fix the issue.
1) ("pci: acpihp: assign BSEL only to coldplugged bridges")
Signed-off-by: Igor Mammedov <imammedo@redhat.com>
Message-Id: <20230112140312.3096331-13-imammedo@redhat.com> Reviewed-by: Michael S. Tsirkin <mst@redhat.com> Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Igor Mammedov [Thu, 12 Jan 2023 14:02:43 +0000 (15:02 +0100)]
pci: acpihp: assign BSEL only to coldplugged bridges
ACPI PCI hotplug would broken after bridge hotplug and then migration
if hotplugged bridge were specified on target at command line.
Currently it's not possible since, 'hotplugged' property was made
read-only for some time now.
The issue would happen due to BSEL being assigned to all bridges
during 1st 'reset':
source seq:
1. start 'pc' machine => sets BSEL to 0 on pci.0 (host-bridge)
2. hotplug bridge, no bsel is assigned (so far is ok)
target seq:
1. start 'pc' machine with
-S -device pci-bridge,id=hp_br,hotplugged=on
BSEL gets assigned to as follows
hp_br: 0
pci.0: 1
as result hotplug requests with migrated AML generated on source
would be misdirected to 'hp_br' instead of intended pci.0
While it's not issue at the moment, it's based on implicit assumptions
* 'hotplugged' property is read-only
* 1st reset happens before QEMU drops into monitor mode
which lets add hotplugged on source bridges as hotplugged ones
(anything added at that stage counts as hotplugged
(yet another assumption))
All of it looks too fragile to me, so lets restrict BSEL only
to cold-plugged bridges explicitly.
Migration wise it shouldn't break anything since assignment order
stays the same:
* user can't specify 'hotplugged=on' on CLI
* user can't specify 'hotplugged=off' at monitor stage or later
on older QEMU versions where 'hotplugged' is RW, hotplug is broken
after migration anyways and we cannot do anything to fix that.
Signed-off-by: Igor Mammedov <imammedo@redhat.com>
Message-Id: <20230112140312.3096331-12-imammedo@redhat.com> Reviewed-by: Michael S. Tsirkin <mst@redhat.com> Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Igor Mammedov [Thu, 12 Jan 2023 14:02:42 +0000 (15:02 +0100)]
pcihp: piix4: do not call acpi_pcihp_reset() when ACPI PCI hotplug is disabled
piix4_pm_reset() is calling acpi_pcihp_reset() when ACPI PCI hotplug
is disabled, which leads to assigning BSEL properties to bridges on path
acpi_set_bsel()
...
if (qbus_is_hotpluggable(BUS(bus))) {
// above happens to be true by default (though it's SHPC hotplug handler)
// set BSEL
}
At the moment the issue is masked by the fact that we use not only BSEL,
to decide if we should generated hoplug AML but also pcihp_bridge_en knob.
However the later patches will drop dependency on pcihp_bridge_en,
and use only BSEL exclusively to decide if hotplug AML for slots should be built,
which exposes issue.
We should not ever call acpi_pcihp_reset() if ACPI PCI hotplug is disabled,
make it so.
PS:
* Q35 does the right thing (i.e. it calls acpi_pcihp_reset only when pcihp is enabled)
* the issue also makes acpi_pcihp_update() logic run on SHPC enabled bridges,
which seems to be harmless
Fixes: 3d7e78aa777 ("Introduce a new flag for i440fx to disable PCI hotplug on the root bus") Signed-off-by: Igor Mammedov <imammedo@redhat.com>
Message-Id: <20230112140312.3096331-11-imammedo@redhat.com> Reviewed-by: Michael S. Tsirkin <mst@redhat.com> Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
projects / qemu-xen.git / log